Prospective, randomized, multicenter clinical trial evaluating longitudinal changes in brain function and microstructure in first-episode schizophrenia patients treated with long-acting injectable paliperidone palmitate versus oral antipsychotics

Widespread anatomical alterations and abnormal functional connectivity have shown strong association with symptom severity in first-episode schizophrenia (FES) patients. Second-generation antipsychotic treatment might slow disease progression and possibly modify the cerebral plasticity in FES patients. However, whether a long- acting injectable antipsychotic (paliperidone palmitate [PP]), available in monthly and every-3-months formulations, is more effective than oral antipsychotics (OAP) in improving cerebral organization has been unclear. Therefore, in the current longitudinal study, we evaluated the differences in functional and microstructural changes of 68 FES patients in a randomized clinical trial of PP vs OAP. When compared to OAP treatment, PP treatment was more effective in decreasing abnormally high fronto-temporal and thalamo-temporal connectivity, as well as increasing fronto-sensorimotor and thalamo-insular connectivity. Consistent with previous studies, multiple white matter pathways showed larger changes in fractional anisotropy (FA) and mean diffusivity (MD) in response to PP compared with OAP treatment. These findings suggest that PP treatment might reduce regional abnormalities and improve cerebral connectivity networks compared with OAP treatment, and identified changes that may serve as reliable imaging biomarkers associated with medication treatment efficacy.


Introduction
Schizophrenia is a complex psychiatric disorder characterized by psychosis, social withdrawal, cognitive deficits, and emotional impairment (Kahn and Giannopoulou, 2015) that affects approximately 0.28 % of the global population (Charlson et al., 2018). Advanced neuroimaging modalities and analytical approaches have contributed to the investigation of this disease (Chari et al., 2019;Chwa et al., 2020;Fan et al., 2019;Goldman et al., 2009;Gur et al., 1999;Honea et al., 2005;Rimol et al., 2012;van Erp et al., 2016;Zhou et al., 2015). Numerous laboratory and clinical studies have demonstrated adverse cerebral alterations in first episode schizophrenia (FES) patients ranging from morphological volume reduction to functional dysconnectivity, and over time the extent of these changes is correlated with the degree of symptom severity. Specifically, significantly reduced prefrontal-thalamic connectivity and increased motor/somatosensory-thalamic connectivity were observed in early stage and chronic schizophrenia and were associated with cognitive functioning (Woodward and Heckers, 2016;Woodward et al., 2012).
Second-generation antipsychotic treatment in FES often modulates brain activity within the thalamocortical network (Glahn et al., 2008;Minzenberg et al., 2009;Shenton et al., 2001;Welsh et al., 2010), resulting in stabilization of disease symptoms as well as brain structure and function. For example, increased activity in the anterior cingulate, bilateral prefrontal and parietal cortices, as well as decreased activity in the putamen and motor network were observed following antipsychotic treatments in SCZ patients (Abbott et al., 2013;Ahmed et al., 2015;Budenholzer, 2019;Sarpal et al., 2015;Scherk and Falkai, 2006;Szeszko et al., 2014;Wang et al., 2017). Additionally, some studies have suggested the use of antipsychotics may be associated with complex changes in white matter integrity, both positive and negative, as measured with diffusion tensor imaging (DTI). For example, studies have shown reductions in fractional anisotropy (FA) following antipsychotic treatment within specific brain areas, which appears to be counterintuitively associated with a more favorable outcome (Szeszko et al., 2014;Xiao et al., 2018). It has also been suggested that, compared to oral risperidone, long-acting injectable formulations may improve the trajectory of intracortical myelination in FES and have a beneficial impact on cognitive performance (Bartzokis et al., 2011;Bartzokis et al., 2012). To date, no studies have compared longitudinal changes in brain microstructure and function between the two different formulations of second-generation antipsychotic medications, long-acting injectable antipsychotics (e.g. paliperidone palmitate [PP]) and oral antipsychotics (OAP), particularly in FES within the context of a randomized clinical trial.
While the mechanism of action for both long-acting and oral antipsychotics are similar, there are distinct differences between these two formulations including differences in pharmacokinetics and potential differences with medication adherence, both of which may cause differences in brain function and microstructure in FES patients. Specifically, we theorize long-acting formulations of antipsychotic medications will exhibit greater functional and structural reorganization in similar areas of the brain but with a higher magnitude than oral antipsychotics, including within and/or between the thalamus, frontal and prefrontal lobe regions, and the temporal lobe. To test this hypothesis, we will utilize the Disease Recovery Evaluation and Modification (DREaM) study (Alphs et al., 2022), a randomized, open-label, delayedstart trial originally designed to compare the effectiveness of paliperidone palmitate (PP) versus oral antipsychotics (OAP) in delaying time to first treatment failure (TtFTF) in participants with recent-onset schizophrenia or schizophreniform disorder.

Patients
The Disease Recovery Evaluation and Modification (DREaM) study (Alphs et al., 2021) was a prospective, open-label, matched-control, multicenter, randomized clinical trial of long-acting injectable vs. oral antipsychotic medication conducted at 34 sites in the United States, Brazil, and Mexico (NCT02431702). Three of the locations that had comparable Siemens 3.0 T MRI scanners (University of California Los Angeles, University of Pennsylvania, and Instituto Nacional de Neurología y Neurocirugía, Mexico City, Mexico) conducted the neuroimaging component of the DREaM study. A total of 68 patients with a recent onset of schizophrenia or schizophreniform disorder were prospectively enrolled in the optional MRI component of the study. The cohort included 57 males and 11 females, ranging in age from 18 to 34 years old, with an average age of 22.6 years (SD = 3.7). The patients had a DSM-5 diagnosis of schizophrenia or schizophreniform disorder, which was established using the Structured Diagnostic Interview for DSM-5 (SCID), and they had experienced a first psychotic episode within the prior 24 months. Detailed demographic and clinical data for the patients are summarized in Table 1. The patients had been stabilized on oral paliperidone ER or oral risperidone antipsychotic medication for 2 months after project entry and were no longer acutely psychotic. All patients provided informed consent and signed Institutional Review Board approved consent forms, and all analyses were performed in compliance with the Health Insurance Portability and Accountability Act (HIPAA). We also recruited 63 age-matched healthy controls (HCs) through community advertisements. The HC cohort had a mean age of 23.0 years (SD = 2.9) and including 46 males and 17 females. HCs had no evidence of a known psychiatric disorder, neurological disorder (e.g., epilepsy) or significant head injury, and did not have a first degree relative who has schizophrenia, schizophreniform, schizoaffective, or bipolar disorder. An exclusion criterion in both participant groups was meeting the DSM-5 definition of moderate or severe substance use disorder (except for nicotine) within 2 months prior to screening.

Study design
Before the first randomization, all patients were prescribed oral paliperidone ER or oral risperidone for 2 months prior to receiving their first MRI scans. The first randomization was 1:2 to either paliperidone palmitate or continued oral antipsychotic. 45 patients got MRI scans 3 months and 9 months following the first randomization. The second randomization was initialized 9 months after the first randomization. At the second randomization, the oral antipsychotic group was randomized 1:1 to either paliperidone palmitate or continued oral antipsychotic. The group on paliperidone palmitate in the first randomization continued on that long-acting injectable antipsychotic medication. Follow-up MRI scans were completed on 20 patients at 3 months and 9 months after the second randomization. For patients randomized to paliperidone palmitate, the one-month formulation was given for a minimum of 4 months followed by the 3-month formulation. The study design for this randomized, multicenter clinical trial is illustrated in Fig. 1. Due to the limited sample size for MRI scans after the second randomization, our analyses focus on the results of the first 9-month randomization period.

Cognitive assessment
The cognitive deficits were evaluated using the MATRICS Consensus Cognitive Battery (MCCB) (Nuechterlein et al., 2008). The MCCB was assessed by a qualified rater at baseline and after 9 months of antipsychotic treatments. Raw scores were converted to normalized T-scores, corrected for age and gender, resulting in an overall composite T-score via the MCCB Computer Scoring Program (Table 1).

MR imaging acquisition
All images were acquired using Siemens 3 T MRI scanners (Prisma or Skyra, Siemens Healthcare, Erlangen, Germany). Resting-state fMRI was acquired with eyes open and a repetition time (TR) = 2000-2090 ms; echo time (TE) = 28 ms; slice thickness = 4 mm with no interslice gap; field-of-view (FOV) = 220 × 220 with an acquisition matrix of 64 × 64 and a flip angle = 77 o . DTI was acquired in 64 directions using b = 1000 s/mm 2 along with a single b = 0 s/mm 2 image using a TR = 8000-10,000 ms; TE = 80-86 ms; flip angle = 90 o ; and FOV = 256 × 256 with an acquisition matrix of 128 × 128 for a 2 mm isotropic voxel size (number of participants and the acquisition parameters for DTI at different scanning sites were summarized in Table S1). A 1 mm 3D isotropic MPRAGE sequence was acquired for alignment with functional and diffusion MRI data using the following parameters: TR = 2300 ms, TE = 2.03 ms, inversion time (TI) = 900 ms, flip angle = 9 o , FOV = 256 × 256mm, matrix size of 256 × 256, and a slice thickness = 1 mm. All neuroimaging data were quality controlled in real time and archived according to multi-site imaging procedures.

Image preprocessing 2.5.1. Functional imaging
The CONN toolbox (https://www.nitrc.org/projects/conn) was used for functional connectivity analysis of the brain (Whitfield-Gabrieli and Nieto-Castanon, 2012), which implements functions from the Statistic Parametric Mapping (SPM, http://www.fil.ion.ucl.ac.uk/spm/) toolbox. All functional MR images were preprocessed using the standard built-in preprocessing pipeline within the CONN toolbox, including: functional realignment (motion correction, 12 degrees of freedom), unwarping, slice timing correction, outlier detection, registration of functional data to the structural volume, registration of the structural volume to the standardized space defined by the Montreal Neurological Institute (MNI), and segmentation of structural volumes, which included skull stripping and processing of tissue types (gray matter (GM), white matter (WM), and cerebral spinal fluid (CSF)). Spatial smoothing of the functional data was performed using an 8 mm FWHM Gaussian kernel. For denoising, signal from the WM, CSF, and motion parameters were regressed from the functional data. Additional signal filtering was performed using a band-pass filter of 0.008 -infinity Hz to reduce noise due to physiological effects, such as respiration and pulsation, and noise due to scanner drift. Lastly, ROI-to-ROI functional connectivity matrices were computed between all brain regions defined in MNI standard space using the CONN toolbox.

Microstructural imaging
All diffusion-weighted images were first denoised using the MRtrix3 software package (Brain Research Institute, Melbourne, Australia, http://www.brain.org.au/software/mrtrix) (Calamante et al., 2012). Then FMRIB's Diffusion Toolbox (FDT) (www.fmrib.ox.ac.uk/fsl) was used to correct for head motion and eddy current geometric distortions, and to affinely register images to the first no-diffusion weighted (b = 0 s/ mm 2 ) image of each participant. Following skull extraction with Brain Extraction Tool (BET), fractional anisotropy (FA) and mean diffusivity (MD) values at the voxel level were calculated for DTI. All FA and MD images were registered to the Johns Hopkins University DTI atlas (ICBM-DTI-81 1 mm FA atlas, Table S2) using the FLIRT and FNIRT commands in FSL. Probabilistic tractography using the MRtrix3 software package was performed to calculate the fiber population per voxel, or fixel (Raffelt et al., 2017). A total of 10 million streamlines, which reflect white matter pathways of the brain, were generated throughout the brain using a step length of 0.5 mm and a maximum of 2000 steps. To provide more biologically meaningful estimates of microstructural connection density, the tractogram (Sporns et al., 2005), which is the whole set of cerebral streamlines, was generated by filtering the overall streamline count to 1 million using the SIFT (spherical-deconvolution informed filtering of tractograms) algorithm (Calamante et al., 2012). Since tractography was performed in native diffusion space, parcellations and segmentations were registered from T1-weighted images to native diffusion space to define the regions between which connectivity was assessed.

Cerebral multiparametric connectivity network construction
Two cerebral networks were constructed according to connection weight: 1) the functional connectivity (FC) network, where the magnitude of the correlation coefficients represent the weights of the specific region in the FC network; and 2) the microstructural connectivity (SC) network, where the number of streamlines represent the weight of the specific regions in the SC network. Seed and target ROIs were selected from the Harvard-Oxford atlas based on previous studies that have observed association between functional and microstructural changes in FES patients, including brain regions in bilateral prefrontal and frontal cortices, bilateral pre-and postcentral gyri, bilateral inferior, middle and superior temporal cortices, multiple regions located in parietal and occipital cortices, bilateral parahippocampus and hippocampus, posterior cingulate, insular cortex, thalamus and basal ganglia, bilateral amygdala, bilateral accumbens, and bilateral cerebellum (Honea et al., 2005;Karlsgodt et al., 2010;Li et al., 2019;Tamnes and Agartz, 2016;Vita et al., 2012;Yu et al., 2012).

Statistical analyses
Following image preprocessing and network construction, a multivariable general linear model (GLM), with age, sex, and site included as covariates, was implemented to identify microstructural and functional alterations within the brain networks associated with the treatment conditions in FES patients. Specifically, we examined 1) differences in FA, MD, functional and microstructural connectivity between FES and HC, 2) longitudinal changes in FA, MD, functional and microstructural connectivity of FES patients who were randomized to PP vs. OAP after 9 months treatments, and 3) associations between identified longitudinal neuroimaging changes and cognitive changes (evaluated by the MCCB overall composite score, controlling for mean daily dosage). For each FA and MD voxel-wise whole brain comparisons, we created GLM through AFNI (Analysis of Functional NeuroImages, https://afni.nimh.nih.gov/) 3dttest++ command, with age and sex included as covariates. For permutation testing used to estimate the proper cluster threshold, we implemented a family-wise error (FWE) correction approach at the cluster level through command 3dClustSim in AFNI. The cluster-extent threshold corresponded to the statistical probability (α = 0.05, or 5 % chance) of identifying a random noise cluster at a predefined voxel-wise threshold of p < 0.05 (uncorrected). For each functional and microstructural connectivity network comparisons, we performed ROI-to-ROI analyses through GLM, with age and sex included as covariates. The level of significance for all statistical analyses was set at p < 0.05.

Functional and microstructural comparisons between FES and HC
By seeding the prefrontal and frontal lobes, we were able to observe that in FES patients at baseline there was stronger FC with the temporal lobe and weaker FC with the somatosensory regions relative to HCs. Seeding the thalamus allowed us to observe stronger FC with the frontal and parietal lobes, but weaker thalamo-insula FC when comparing to HCs ( Fig. 2A). White matter tractography further identified stronger SC from the frontal lobe to the temporal lobe and areas within the limbic system in FES patients (Fig. 3A).
Widespread microstructural abnormalities, such as decreased FA within supratentorial regions and decreased MD values within infratentorial regions, were also present in the brains of FES patients ( Fig. 4A and Table 2). Additionally, in FES patients, the corpus callosum showed decreased FA and increased MD, while the superior corona radiata showed increased FA and decreased MD ( Fig. 4B and Table 3).

Cognitive improvement within FES patients after 9 months of OAP or PP treatments
Prior to any treatment (at baseline), the MCCB overall composite score of FES patients who were going to receive OAP treatment was 35.4, while the MCCB overall composite score of FES patients who were going to receive PP treatment was 35.3. After nine months of treatment (phase II), cognitive improvement was observed in both the OAP group (p = 0.0143) and the PP group (p = 0.0615). the MCCB overall composite score of FES patients who received 9-month of OAP treatment was 38.5, while the MCCB overall composite score of FES patients who received 9-month of PP treatment was 39.9. (Fig. 5).

FC and SC changes within FES patients after 9 months of OAP or PP treatments
No significant difference in age (p = 0.56) was identified between OAP and PP groups. When seeding the prefrontal and frontal lobes in FES patients, both OAP and PP treatment resulted in an increased FC to somatosensory regions. Additionally, increased fronto-temporal FC was observed in the OAP group but decreased fronto-temporal FC was more pronounced in the PP group, which partially normalized the abnormal FC noted at baseline relative to HCs (Fig. 2B). When seeding the thalamus, increased FC to the temporal lobe and decreased FC to the frontal lobe was observed in FES patients receiving OAP treatment but decreased FC to the temporal and frontal lobes was observed in FES patients receiving PP treatment (Fig. 2B). We also observed increased FC from the thalamus to the insula and the anterior cingulate gyrus following PP treatment. When examining the association between the cognitive improvement and FCs changes, most of the FC changes were negatively correlated with MCCB overall composite score changes in both OAP and PP groups, except the fronto-parietal FCs and thalamofrontal FCs showed positive association with improved MCCB overall composite score in FES patients receiving PP treatment (Fig. S3).
When examining the SC changes (Fig. 3B), both increased and decreased SC to other cortices were observed in the OAP group by seeding the thalamus, prefrontal and frontal regions. In contrast, decreased SC among multiple cortical regions was seen in the PP group. Specifically, abnormal SC observed in FES patients when compared to HCs at baseline, including stronger fronto-temporal and thalamo-frontal SC, were significantly reduced after 9 months of PP treatment. In addition, when seeding the thalamus, prefrontal and frontal regions, we found decreased SC to the parietal lobe in the PP group. When examining the association between the cognitive improvement and SC changes, most of the SC changes were negatively correlated with MCCB overall composite score changes in both OAP and PP groups, except the anterior-posterior cingulate SCs, fronto-temporal SCs, and thalamofrontal SCs, showed positive association with improved MCCB overall composite score in FES patients receiving PP treatment (Fig. S4).

FA and MD changes within FES patients after 9 months of OAP or PP treatments
No significant FA changes were observed after nine months of OAP treatment, while after nine months of PP treatment ( Fig. S1 and Table S3) patients showed a reduction in FA values along white matter tracts connecting temporal and frontal lobes ( Fig. 6 and Table S4). Four localized clusters within the left cingulate gyrus, the body of corpus callosum, the thalamus, the right internal and external capsules, as well as along fiber bundles including the right superior corona radiata and the right uncinate fasciculus, demonstrated decreased FA after nine months of PP treatment. Additionally, three of the identified clusters showed positive association between FA changes and MCCB overall composite score changes in FES patients who received 9-month of PP treatment ( Fig. 6C and Table S4).
Increased MD was observed in both groups following nine months of antipsychotic treatment (Fig. S2, and Tables S5-S6). Compared to OAP treatment, after nine months of PP treatment, five localized clusters within the left cingulate gyrus, the right internal capsule, the thalamus, the body and the splenium of corpus callosum, as well as along fiber bundles including the right corona radiata, the right superior frontooccipital fasciculus and the bilateral superior longitudinal fasciculi, demonstrated greater increased MD (Fig. 7 and Table S7). Additionally, two of the identified clusters showed negative association between MD changes and MCCB overall composite score changes in FES patients who received 9-month of PP treatment (Fig. 7C and Table S7).

Discussion
The current study investigated temporal changes in the brain in FES patients associated with the long-acting injectable antipsychotic, paliperidone palmitate, as compared to oral antipsychotics, by integrating microstructural and functional imaging measurements. We demonstrated abnormal functional and microstructural connectivity within and from the prefrontal/frontal lobes, thalamus and basal ganglia in FES patients relative to HCs. In addition, widespread white matter abnormalities, including altered FA and MD values within internal capsule and along multiple projection and association fibers, were also observed in the brains of FES patients when comparing to HCs. Using whole brain voxel-wise analyses and ROI-to-ROI analyses by seeding specific regions in the prefrontal and frontal lobes and the thalamus   known to be involved in the pathophysiology of FES, results demonstrate that treatment with long-acting injectable PP might reorganize brain microstructure and function in patients with FES beyond that observed with treatment with OAP. Studies have consistently reported localized changes in the prefrontal/frontal cortices and thalamus in FES patients, which may serve as an imaging biomarker to monitor disease progression and evaluate treatment efficacy (de Leeuw et al., 2015;Li et al., 2019;Littow et al., 2015;Suzuki et al., 2005;Zhang et al., 2014;Zhou et al., 2015). The prefrontal/frontal cortices play an essential role in the organization and control of goal-directed thought and behavior (Szczepanski and Knight, 2014), while the thalamus is connected to all cortical areas and conveys information to the neocortex, including information from the basal ganglia. Abnormal connectivity within and from those regions has been used to characterize the onset of schizophrenia, including reduced connectivity from the prefrontal/frontal cortices to the thalamus and primary sensorimotor regions, increased connectivity from the prefrontal/frontal cortices to the temporal lobes, as well as increased thalamo-temporal and thalamo-sensorimotor connectivity (Giraldo-Chica et al., 2018;Li et al., 2017;Walther et al., 2017;Zhou et al., 2015). Consistent with those previous studies, abnormal functional and microstructural connectivity was observed within and from the prefrontal/frontal lobes and thalamus in FES patients relative to HCs at baseline, and significant functional and microstructural reorganization was observed in FES patients with PP or OAP treatment. Specifically, 9 months of PP treatment increased the FC between the prefrontal/frontal lobes and somatosensory regions, but decreased both the FC and SC to the temporal lobe from the thalamus and frontal lobe, while 9 months of OAP treatment increased both the FC and SC to the temporal lobe from the thalamus and frontal lobe. Those results indicated that PP treatment might be more effective in correcting abnormal aspects of brain connectivity than OAP, while both produce some organizational changes over time.
Previous studies reported increased FC between striatal regions and limbic regions such as the anterior insula following antipsychotic treatment, which influences salience processing as psychotic symptoms are reduced (Sarpal et al., 2015;Szeszko et al., 2014). Seeding the thalamus, we observed weaker thalamo-insula FC in FES patients compared to HCs, and increased FC between the bilateral insula, the anterior cingulate gyrus and the left thalamus was identified in response to 9 months of PP treatment but not OAP treatment.
Numerous longitudinal MRI studies have investigated effects of antipsychotic treatment on brain microstructure (Borgwardt et al., 2009;Chwa et al., 2020;Kraguljac et al., 2016;Scherk and Falkai, 2006;Tishler et al., 2018). One of the most common findings has been FA reductions following antipsychotic treatment within the parietal and occipital lobes and the bilateral anterior cingulate gyri, as well as along the right anterior corona radiata of the frontal lobe (Kraguljac et al., 2019;Meng et al., 2019;Szeszko et al., 2014;Wang et al., 2013). Consistent with those observations, we found that PP treatment was more potent than OAP treatment in decreasing FA and increasing MD within the cingulate gyrus, the body of the corpus callosum, and internal and external capsules, as well as along fiber bundles such as the superior corona radiata, which were observed to be abnormal in FES patients at baseline when compared to HCs. While this would appear to indicate that treatment with PP may accelerate certain aspects of the disease trajectory compared to OAP, as studies of long-term untreated schizophrenia patients have shown dramatically reduced FA compared to treated patients (Xiao et al., 2018), the implications of this on the status of the disease are not straightforward. For example, a clinical trial by Szeszko et al. (2014) showed that a larger reduction in FA was associated with elevated levels of low density lipoprotein (LDL) in blood serum, and a higher level of LDL has been shown to be correlated with response to atypical antipsychotic drugs in schizophrenia patients (Huang and Chen, 2005) including FES (Gjerde et al., 2018). Thus, it is conceivable that a higher reduction in FA observed in FES patients treated with PP may suggest a favorable antipsychotic response and may have resulted in higher serum LDL. The reductions in FA may thereby represent compensatory mechanisms in brain reorganization that are associated with consistent second-generation antipsychotic adherence.
Both OAP and PP groups tended to show cognitive improvement after 9 months of treatment. We demonstrated that both FC and SC changes were negatively correlated with MCCB overall composite score changes in many regions in both OAP and PP groups, suggesting that reduced FC and SC in many regions facilitate cognitive improvement. Beyond these negative correlations, we found that selected FC and SC increases were associated with improved cognitive performance in the PP treatment group, indicating that PP promoted these beneficial connectivity gains. The current study also found that the cognitive improvement significantly correlated with increased FA and decreased MD (within identified clusters showing treatment differences) in FES patients receiving PP, suggesting that PP treatment might be play a therapeutic role in promoting white matter myelination (Bartzokis et al., 2011;Walterfang et al., 2011).
Lastly, limitations of the current study should be noted. Although a reasonably large cohort of FES patients were involved at the baseline MRI in this prospective clinical trial, this was only a subset of the total patients enrolled in the larger clinical trial (68 of 235, or ~ 29 %) and an even fewer patients in each treatment group continued until the end of second randomization. This led to our decision to examine the MRI data only from the first randomization period. Despite our focus on this first period of randomization, we still observed significant dropout even after 9 months of follow up in both treatment groups in the study. While some dropout was expected due to the fact that medication non-adherence is  Fig. 6. A) Regions demonstrating changes in fractional anisotropy (FA) between oral antipsychotics (OAP) and a long-acting injectable antipsychotic (PP) treatments at 9 months in FES patients. Red-Yellow denotes decreased FA in patients receiving PP treatment, while Blue-Light Blue denotes decreased FA in patients receiving OAP treatment. B) Clusters demonstrating changes in FA between OAP and PP treatments at 9 Months in FES patients. C) Clusters demonstrating associations between FA changes and MCCB overall composite score changes in FES patients who received the long-acting injectable antipsychotic (paliperidone palmitate [PP]) treatment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 7. A) Regions demonstrating changes in mean diffusivity (MD) between oral antipsychotics (OAP) and a long-acting injectable antipsychotic (PP) treatment at 9 months in FES patients. Red-Yellow denotes increased MD in patients receiving OAP treatment, while Blue-Light Blue denotes increased MD in patients receiving PP treatment. B) Clusters demonstrating changes in MD between OAP and PP Treatments at 9 Months in FES patients. C) Clusters demonstrating associations between MC changes and MCCB overall composite score changes in FES patients who received the long-acting injectable antipsychotic (paliperidone palmitate [PP]) treatment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) common with oral antipsychotic medication, additional dropout was observed due to a number of factors including need for specialized transportation and logistical arrangements to dedicated imaging centers as well as claustrophobia and anxiety associated with brain MRI exams. Therefore, a larger longitudinal collection of scans from FES patients is necessary to confirm and verify the observations outlined in the current study. Additionally, medication exposure prior to the first randomization, as well as differences in equivalent dose, may impact FC and SC measurements within each respective treatment group. Future studies modeling the changes in FC and SC in response to changes in antipsychotic equivalent dose independent of formulation are warranted to better understand potential dose-dependent brain alterations. Lastly, the effects of sex on disease trajectory and therapeutic response were not considered due to a limited sample of female participants, despite evidence males and females have differing cognitive performance (Goldstein et al., 1998;Lewine et al., 1996;Perlick et al., 1992) and response to treatment (Grigoriadis and Seeman, 2002).

Conclusions
Results from the current analyses suggest that treatment with longacting injectable antipsychotics might reduce regional abnormalities and abnormal brain connectivity more than treatment with oral antipsychotics in the early course of schizophrenia. Identified changes might serve as reliable imaging biomarkers to aid evaluation of antipsychotic medication efficacy and monitoring of treatment progress.

Role of the funding source
This study was funded by Janssen Scientific Affairs, LLC, Titusville, NJ. The study sponsor was involved in the design and conduct of the study; collection, management, and analysis of data; and review and approval of the manuscript. Analyses and interpretation of the MRI data in this manuscript were completed by the UCLA authors. All authors had full access to the study data and take responsibility for data integrity and the accuracy of the analyses. All authors reviewed and approved the final version prior to submission.

Contributors
Wang, Tishler, Nuechterlein, and Ellingson designed the study and wrote the protocol. Wang and Ellingson performed literature searches. Wang performed technical and statistical analyses, and drafted the manuscript. Tishler, and Oughourlian, Nuechterlein, de la Fuente-Sandoval, and Ellingson revised the manuscript. All authors contributed to and have approved the final manuscript.

Declaration of competing interest
Keith Nuechterlein has received research grant support from Posit Science, Inc., Janssen, and Alkermes and has been a consultant to Astellas, Genentech, Janssen, Medincell, Otsuka, Takeda, and Teva. Benjamin Ellingson has received research support, is a paid consultant, and/or is a scientific advisory board member for MedQIA, Agios, Siemens, Janssen, Medicenna, the National Institutes of Health, Imaging Endpoints, Novogen, and Northwest Biopharmaceuticals. Camilo de la Fuente-Sandoval has received research support from Janssen. The other authors report no conflicts of interest.